不规则机票表格的高精确度多单元字符定位
High Accuracy Locating Multi-fields Character Strings in Irregular Airline Coupon Image
-
摘要: 提出一个检测和定位机票图像中污损字符串的技术.不规则机票表格中的字符串是由计算机打印生成的,其对比度低且容易偏离表格,无法用常用的方法处理.基于几何先验知识和字符提取结果,应用扩展最大平均相关高度相关滤波器检测,进而准确地定位期望字符区域.实验结果表明:该方法准确率高,可以为OCR提供良好的输入.Abstract: A technique to detect and locate smudgy character strings in complex tabular coupon images is presented.In our researchcharacter strings in the irregular coupon images are computer-printed but of low contrast and could be shifted out of the table so that they are unable to be detected and located using conventional algorithms.Based on geometric information and string retrieval result extended maximum average correlation height (EMACH) correlation filter is applied as an adaptive shift locator to detect and locate expected character strings.The results demonstrate that the algorithm is of high accuracy and provides excellent input for OCR.
下载: